opencode-workflow/skills/analyze-design-inputs/SKILL.md

3.1 KiB

name description
analyze-design-inputs Extract implementation surfaces and sequencing constraints from architecture and code design documents. Planning pipeline step one. Produces internal analysis only and does not write files.

This skill analyzes upstream design inputs before planning begins.

Announce at start: "I'm using the analyze-design-inputs skill to extract implementation surfaces and sequencing constraints from the design documents."

Purpose

Read the architecture and code design documents together, identify all implementation surfaces, and determine the natural task boundaries and dependency constraints for planning.

Important

This skill produces internal analysis only. It MUST NOT write any file artifacts.

Hard Gate

Do NOT generate tasks in this skill. This skill only extracts, classifies, and organizes planning inputs. Task generation happens in generate-plan.

Process

You MUST complete these steps in order:

  1. Read the architecture document end-to-end
  2. Read the code design document end-to-end
  3. Extract all implementation surfaces:
    • schema and migration surfaces
    • packages/modules to create or update
    • interfaces and model definitions
    • handlers/transports/endpoints
    • services/use cases
    • repositories/data access boundaries
    • async producers/consumers
    • external integrations
    • config and DI wiring
    • build/deploy and operational surfaces
  4. Identify sequencing constraints:
    • what must happen before implementation can start
    • what depends on migrations or schema rollout
    • what depends on interface/model definitions
    • what can be parallelized
  5. Identify risk-heavy work:
    • backfills
    • dual-write/read transitions
    • zero-downtime rollouts
    • external dependency setup
  6. Identify potential task groupings that would make good milestones
  7. Flag upstream ambiguities that block planning

Internal Analysis Format

Retain this analysis internally. Do not write it to disk.

## Inputs Reviewed
- Architecture: `docs/architecture/{feature}.md`
- Code Design: `docs/code-design/{feature}.md`

## Implementation Surfaces
| Surface Type | Item | Source Doc | Planning Notes |
|--------------|------|------------|----------------|
| migration | ... | architecture/code-design | ... |
| repository | ... | code-design | ... |
| service | ... | code-design | ... |

## Dependency Constraints
| Item | Depends On | Reason |
|------|------------|--------|
| ... | ... | ... |

## Parallel Work Opportunities
| Group | Tasks That Can Run In Parallel | Preconditions |
|-------|--------------------------------|---------------|
| ... | ... | ... |

## Risk Notes
| Risk | Affected Work | Planning Impact |
|------|---------------|-----------------|
| ... | ... | ... |

## Upstream Ambiguities
List blockers requiring clarification. If none, write "None identified."

Guardrails

Do:

  • Analyze both upstream inputs together
  • Focus on implementation surfaces and dependency structure
  • Preserve strict separation from design and QA responsibilities

Do not:

  • Create tasks
  • Change design decisions
  • Write the plan file
  • Produce any file artifacts